Upload
khangminh22
View
0
Download
0
Embed Size (px)
Citation preview
A diachronic account of converbal constructions in Old Rajasthani
Abstract
The aim of the paper is to present results of a multilayered diachronic analysis of one of the most
important subordinating devices in IA, i.e. converbs. This is a corpus-oriented study based on an
early variety of NIA, namely Old Rajasthani. We discuss: a) the morphology of the converb and
its possible diachronic pathways of evolution; b) the syntactic properties of converbal chains with
a special focus on main argument marking and the ‘same-subject constraint’ and c) the semantics
of the constructions and the adverbial and clause-chaining properties of converbs. The results of
our research bring interesting implications as regards the aspectual status of the converb which
seems to determine the main argument marking in converbal chains.
Key words: Early New Indo-Aryan, Old Rajasthani, converb, historical morphosyntax
1. Introduction
The term “converb” stems from Altaic linguistics (Ramstedt 1903), as the phenomenon occurs in
many forms in Turkish, Mongolian and other related languages. Only much later the notion of
“converb” was employed in typological linguistics (cf. Haspelmath & König 1995). Since
Haspelmath (1995), an attempt to outline a cross-linguistic, universal definition of “converb”, this
category has been well known in typological linguistics. In order to define the concept,
Haspelmath (1995) lists a number of properties specific to converbs, the most important ones
being non-finiteness, adverbial modification and subordination. In his words, “converbs are
verbal adverbs, just like participles are verbal adjectives” (Haspelmath 1995: 3). However, as for
all universal concepts, it is unclear how many properties of a category should be present to
assume that the category is identical across more than one language. In his review of Haspelmath
(1995), Bickel (1998) is not convinced of the universality of the concept of ‘converb’, but instead
assumes that that there are at least two types of categories subsumed under Haspelmath’s
definition, a European and an Asian one. Moreover, he argues that to define a converb as purely
“adverbial” is sensu stricto too limited; in many cases the converb functions on an “adsentential”
level, “providing a topic or framework for subsequent discourse” (Bickel 1998: 384), rather than
simply modifying a sentence. Furthermore, Haspelmath distinguishes different layers in the
analysis of converbs, i.e. analyses of converbs can/must be performed on a syntactic,
morphological and semantic level. Only by taking into consideration these three levels, we can
hope to find a convincing definition of converbs.
The goal of this study is less ambitious: we aim to describe the development and the features of
converbs in the earliest stages of New Indo-Aryan (NIA). In Indo-Aryan (IA) linguistics, they
were studied from an Indo-European point of view, a classical approach which considers them as
participial forms (hence, the traditional names such as ‘gerund’ (e.g. Tikkanen 1987) or
‘conjunctive participle’ (e.g. Kellogg 1876: 162; Caldwell 1875: 398)). However, in later works
and after noticing that a similar construction also occurs in Dravidian, a more typological
perspective has been taken, as can be found in the works of Davison (1981), Kachru (1981),
Subbarao (2012) etc. Furthermore, converbal constructions are mentioned as a defining feature of
the linguistic area of South Asia (Masica 1976).
Besides describing the synchronic properties of converbs, certain diachronic questions rise about
the development of converbs in languages. Haspelmath (1995: 17) mentions two possible origins
of converbs. The most common origin is the verb combined with a case ending. Typical for the
classical languages such as Latin and Greek, invariant forms of participles were used in a
converbal sense. The second source of converbs can be co-predicative participles which in the
course of time became uninflected forms. However, as Haspelmath (1995: 17) acknowledges
himself, cross-linguistically the origin of converbs needs further investigation. There is a second
diachronic point of interest, i.e., how does the converb further evolve? In particular, according to
Haspelmath (1995: 37), converbs could easily grammaticalize into adpositions1, losing their
“verbal” meaning, but gaining a meaning more related to case. The Hindi converb lekar is such
an example, used in the adpositional sense of ‘with’ instead of the verbal ‘after having taken’.
Historical investigations of the development of converbs in IA languages have thus far not been
undertaken in a systemic, empirical way. This study is a first attempt to give an overview of the
properties of converbs in Old Rajasthani, a language belonging to the IA branch. Following
Bickel, we distinguish the three different layers of syntax, semantics and morphology, to be as
exhaustive as possible.
Our data is a 10000 word annotated corpus of Old Rajasthani short prose texts ranging from the
14th
to the 18th
century (Bhānāvat and Kamal 1997–1998). The corpus has been annotated by
means of IATagger (Jaworski 2015) at the level of morphosyntax and semantics.
1 Consider, for instance, the form ādāya, meaning ‘taking’ in Sanskrit, which has grammaticalized into the
postposition with the meaning ‘with’ in Pali (Bloch 1965:159).
We will also refer occasionally to patterns found in Pahari, Awadhi and Braj, other Early NIA
languages, which were spoken in the same time and environment as Old Rajasthani. For these
languages, we are also developing corpora of the same size.
In the main body of the article, we will often refer to the use of converbs in modern NIA, in
particular Hindi and Rajasthani. The studies available on modern NIA are considered as a point
of departure for historical comparison. The article is structured as follows: in the next section, we
will discuss the attested forms of the converb in Old Rajasthani, and possible diachronic
pathways of evolutions that these forms have followed. Third, the syntactic constructions in
which these converbs occur will be discussed, in particular with reference to the marking of the
arguments they occur with, the so-called “same-subject constraint”, and their use in serial verb
constructions. Fourth, we discuss the semantics of the construction and the adverbial and clause-
chaining properties of converbs. In the last section, we conclude by giving an outline of the
properties of Old Rajasthani converbal forms.
2. Morphology of Old Rajasthani converbs
Converbs are generally non-finite forms, which means that they are considered to be verbal, but
they do not indicate any marker of tense, aspect, mood or agreement. The bare root form is the
formal equivalent of such a non-finite form, since there are indeed no formal indications of any of
these parameters. Among the varieties of converbal forms in modern NIA, simple roots can occur
as converbs in Hindi. Apart from these root forms, Hindi speakers add the suffix –kar (or a
contracted form -ke) in order to form a converb. This seems to go back to the Apabhramśa form
kari (< Skt. kr̥te, locative of the past perfect participle kr̥ta) derived from the verb kar- ‘to do’ (cf.
Oberlies 2005: 44, Reinöhl 2016). In Nepali, we find a more extensive spectrum of converbal
forms, for instance, the form ending on –era, arguably an old participial form (Bickel 1998), and
forms ending on -ī or in -īkana (the last one being emphatic and usually used in negation).
In our study, we looked at a sample of Old Rajasthani texts from the 14th
century onwards. We
found a number of different formations of converbs, given in Table 1.
Table 1 Morpohology of the Early Rajasthani converb
I Ī Ī+ karī i+ karī i+nai i+ara i+ari i+kāri ya ya+ nai root root + nai root + karī root+ara root + kara
dvitīya vrata
satya para
kathā (14c.)
1
1
15 19 1 2
guru mahimā
par kathā
(15c.)
1 10
amarsen-
vayarsen (15c.)
55
vacanikā khīcī
acaḷadāsa-rī
(15c.)
8
dalapata-
vilāsa
(16/17c.)2
1
5
2 9 1 1 3 1 3
hadai
surijamal-ri
bata (17c.)
1 3 2 22 5
vacanika
rathod ratan ri
(17c.)
9 1 1
2The Dalpat Vilas was probably written during Maharaja Rai Singh’s rule in the period 1579–1612 C.E.
rathod
duradavasa ro
kagada
(17/19c/)
1 1
dhanush
bhamg (18c.)
1
7
6 8 6 1 7 1
adalati nyay
(18c.)
12 1 28 3
dokari ri bāt
(18c.)
2
The verb followed by the suffixes -ī and-i is the most frequent converbal form. For instance, in
the following example, the converb dekhī is used:
(1) te dekh-ī vismaya hūntu loka
this see-CVB3 astonished be.PTCP.PRS.M.NOM.SG people.M.NOM.SG
teha-hnaiṃ pūjābhakti karaiṃ.
s/he-ACC worship.F.NOM.SG do.PRS.3PL
‘Having seen it, the astonished people worshipped him.’ 15th
c. (R.C.)
In our Old Rajasthani data, the root of the verb (caḍha in the ex. (2)) was already in use in the
sense of a converb:
(2) tarai rāṇo ghoṛai caḍha sūrajamala=nūṃ
then king.M.NOM.SG horse.M.OBL.SG climb.CVB Surajmal=ACC
jhaṭako vāyo
sword.M.NOM.SG strike.PST.M.SG
‘Then the king, having climbed on his horse, stroke Surajmal with his sword.’ 17th
c
(R.C.)
The origin of the Hindi converbal suffix –kar is also clear from our data, as there are a number of
instances of an –ī and-i form followed by karī or kari, which is, in itself, a converb of ‘to do’.
3The following abbreviations occur in the glosses: ABL: ablative, ACC: accusative, AUX: auxiliary, CVB: converb,
ERG: ergative, F: feminine, FUT: future, GEN: genitive, INF: infinitive, LOC: locative, M: masculine, NOM:
nominative, OBL: oblique, PL: plural, PRS: present, PST: past, PTCP: participle, SBJV: subjunctive, SG: singular,
TR: transitive.
(3) mārgu melhī-karī āghau cāliu
way.M.NOM.SG leave-CVB far go.PST.M.SG
‘Having left the road, he went further on.’ 14th
c. (R.G.)
The suffix –ara is also sometimes added, the origin of it is unclear, but it is perhaps related to
Modern Nepali –era.
(4) pāchai damāmo de-ara caḍhiyo akabara
after drum.M.NOM.SG give-CVB climb.PST.M.SG Akbar.NOM.M.SG
pātisāha dilī-nūṃ
king.NOM.M.SG Delhi-ACC
‘After giving the order to the drums, King Akbar climbed up to Delhi.’ 16/17th
c. (R.G.)
The form followed by a postposition nai is not common anymore in contemporary NIA. nai can
be added to the root of the verb (5), but also to a form extended by a pleonastic suffix -a (cf.
Tessitori 1915: 105), as in ex. (6).
(5) to hūṃ bhāṇā sarīkhā pātra=nai de-nai
then I.NOM.SG Bhana.NOM.SG like poet=ACC give-CVB
amara karūṃ
immortal.NOM.M.SG do.SBJV.1SG
‘Then, after I gave them a poet as Bhana, I shall act as an immortal.’ 17th
c. (R.G.)
(6) ti vārai paga dhoya-nai nadī pāra ūtariyā
this time foot.M.NOM.PL wash-CVB river.F across pass.PST.M.PL
‘Then, after he washed his feet, he passed across the river.’ 18th
c. (R.G.)
It is unclear whether this latter form, exemplified by dhoyanai (from dhoya- ‘wash’) should be
counted as a converb, or as a participial form taking case endings. Semantically though, it has
clearly a different function from a participle, as it adverbially modifies the event expressed in the
main verb. The formation clearly shows the non-finite nature of converbs, and their position
between nominal and verbal forms. We also find forms of verbs ending on –i, followed by nai,
e.g. dekhinai. This formation seems to be purely nominal: the regular ending of a converb in Old
Rajasthani -ī has been interpreted as a locative case ending (Tessitori 1915: 119; see also
Chatterji 1926: 1010-1011) although there has also been another, even more common, hypothesis
deriving the converbal ending from Apabhramśa -i < Skt. -ya (e.g. Jha 1958: 514; Oberlies 2005:
44). By adding the postposition nai4, which is a multifunctional postposition normally used to
indicate (pro)nominal core arguments, one gets a better grip on the function of this postposition
in Old Rajasthani: it functions merely as a reinforcement of the oblique (locative), indicating that
an argument is no longer in the unmarked, nominative case. In these examples in particular, it
indicates that the verb has assumed an adverbial function.
As expected when considering nai as a reinforcement, we notice that in texts up to the 14th
century, the form on –ī and –i are clearly preferred; in later texts such as the Dalpata Vilasa from
the 16th
century, all forms are attested. We can assume that the addition of nai has developed in
parallel with the tendency towards postpositional marking of arguments that followed the
extensive case syncretism in Prakrit. When gradually the function of nai became specified
towards indicating core arguments with a high level of animacy, nai as a marker of a converb
disappeared, until it stopped occurring in Modern Rajasthani.
4 Tessitori (1915: 119) derived nai from the locative kane which is possible interpretation of the origin of this
postpostion but there are also other possible derivations and no one has been proved final (cf. Stroński 2009).
Comparative data from other IA languages clearly shows that the preferred form in early NIA
was the one terminating in -i. This is the form attested in early Awadhi e.g. ā-i ‘having come’
(Jāyasī ‘Padmāvat’1540 A.D.) along the ones terminating in: kai e.g. dekhi kai ‘having seen’
(Jāyasī ‘Padmāvat’ 1540 A.D.; cf. Śukla 1965: 153), kara e.g. jāi kara ‘having gone’ and kari
e.g. khāi kari ‘having eaten’ (Tulsīdās ‘Rāmcāritamānas’ 1575 A.D.; cf. Saksena 1972 [1937]:
281). Early Pahari had a main converbal form terminating in -i e.g. Nepali ghāl-i ‘having put’
(inscription from 1398 A.D.; Pokharel 1963: 32) but the form in –era appeared in Nepali also
quite early – jag-era ‘having saved’ (inscription from 1398 A.D. (Chalise 2006: 267)).
3. Syntactic dimensions of the converbs in Old Rajasthani, with particular emphasis on
the “same-subject constraint”.
Just as for so many other linguistic topics, converbs in Hindi are the best studied in IA. One of
the main points of study is the so-called “same-subject constraint”, according to which the subject
of the converb must be coreferential with the subject of the main verb (see for example Subbarao
2012: 264-282). Most commonly, it follows that the main verb determines the marking of the
subject, not the converb – there is no separate, explicit mention of the subject of the converb. At
first sight, it turns out that this is a valuable constraint for Hindi-Urdu. However, after further
investigations, it appears that pragmatic motivations can also play part (cf. Tikkanen 1995: 496),
and the converbal subject is then coreferential with another (overt) argument, not necessarily the
subject argument. Consider the following examples: example (7) shows a construction in which
the main verb and the converb share the subject rām. Hindi has an ergative case marking pattern
in perfective constructions, but in this example, there is no ergative marking of the subject
because the main verb is intransitive – in contrast to the converb. The example (8) is the opposite
situation, with a transitive main verb, which motivates the ergative marking of the subject rām.
Example (9) is an unlikely construction, since the subject of the converb is not coreferential to
any argument of the main verb. However, a transgression of the same-subject constraint is
possible as well, as ex. (10) and (11, from Bickel and Yadav 2000: 351) show. The difference
between the ungrammatical (9) and the grammatical (10) is of course that the subject of the
converb in the last two constructions is involved as a salient argument in the main clause as well,
though not necessarily being the subject argument. In (11), this is less clear, though it is implied
that there will not be work for you. In that sense, tumhāre is an argument of the main clause as
well. Note that the subject of the converb, tumhāre, is a genitive form, which indicates the
tendency to nominality of the converb. An alternative translation would be “your not coming”.
With regard to the the genitive case of tumhāre, this is a different construction from a dative
subject construction. It is unlikely that tumhāre would be replaced by a dative form in this
particular construction.
(7) rām soc-kar ghar calā gayā
Ram.M.NOM.SG think-CVB home go.PST.M.SG go.AUX.PST.M.SG
‘Ram, having thought, went home’.
(8) rām=ne soc-kar sigret pī
Ram.M.=ERG think-CVB cigarette.F.NOM.SG drink.PST.F.SG
‘Ram, having thought, smoked a cigarette.’
(9) [*rāmi soc-kar] laṛkīj=ne kām kiyā
Ram.M.NOM.SG think-CVB girl.F= ERG work.M.NOM.SG do.PST.M.SG
*‘Ram having thought, the girl did the work.’
(10) yahaṃ nahīṃ ā-kar maiṃ tujhei kuch nahīṃ sikhā
here not come-CVB I.NOM.SG you.OBL.SG nothing teach
saktā+hūṃ
can+be.PRS.1SG
‘If you don’t come here I will not be able to teach you.’
(11) tumhāre kal nahīṃ ā-kar kuch kām nahīṃ hogā
your.GEN.SG tomorrow not come-CVB some work.M. not be.FUT.M.SG
‘If you don’t come tomorrow, there won’t be any work.’
The second language which is often cited with regard to converbs, is Nepali. Of Nepali, it is
generally agreed that the subject of the converb is pragmatically determined. Consider the
following example from Bickel (1998: 394):
(12) timī=le bhan-era mātrai yo kām gar-eṃ
you=ERG say-CVB only this work.M.NOM.SG do-PST.1SG
‘I did this work only because you told me so.’
This example shows the subject of the converb with an ergative marking, whereas the subject of
the main clause is an unexpressed first person. In Wallace’s opinion (1982), converbs in Nepali
are as per definition perfective5; thus, if they are transitive, their subject is marked with the
ergative case. However, in Nepali, in many cases, the subject of the converb is different from that
5 Several scholars claimed the perfectivity of Hindi converb (e.g. Davison 1981: 117-121); others (Kachru 1981)
added to this interpretation meanings such as simultaneity. It seems that the diachronic evidence from early NIA
rather confirms the predominant perfective status of the converb (Stroński and Tokaj 2015) but certainly more
research has to be done in this respect.
of the main verb. Hence, there is no conjunction reduction and both subjects are marked
according to the transitivity of their controlling verbs. What is remarkable though, is that when
both the main verb and the converb share the same subject, its case marking is either determined
by the transitivity of the main verb or by the converb. Both options are possible. Below are two
of Wallace’s examples (1982: 168) in which the converb is transitive and the main verb
intransitive, and both times, the converb determines the ergative marking of the subject:
(13) mai=le tyas=lāi bheṭ-era ghar=mā ga-eṃ
I=ERG he=ACC meet-CVB house=in go-PST.1SG
‘I met him and went home.’
(14) us=le bikh khā-era mar-yo
he=ERG poison eat-CVB die-PST.3SG
‘He died while eating poison.’
Nevertheless, the opposite pattern is more frequent, when the main verb determines the case
marking of the subject, as illustrated in the next example (Hutt 1997: 49).
(15) unī korībātī gar-era jhyāl=mā ga-era daṭ-in
she make-up do-CVB window=in go-CVB stand-PST.F3SG
‘After she put on her make-up, she went standing in the window.’
The same observations have been made for Assamese (Verbeke 2013: 116-117). Consider the
following example (Baruah 1980: 865), in which the subject of the converb takes the ergative
ending –e.
(16) mīnuw-e chabi āṃkiba-lai kal-at si chabi āṃkiba-lai dhar-il-e
Minu-ERG picture draw-ACC ask-CVB he picture draw-ACC begin-PST-TR.3SG
‘After Minu asked to draw a picture, he started to draw a picture.’
Both Assamese and Nepali show a spread of ergative marking towards imperfective
constructions, which perhaps underscores their pragmatic approach towards subject marking.
Note that it also seems to be a rule in Assamese that the marking of the subject is determined by
the verb which is linearly the closest to the subject (Verbeke 2013: 117).
In contrast to Nepali and Assamese, Old Rajasthani generally obeys the same-subject constraint,
and the marking of the shared argument is decided by the main verb. Consider the following
examples. In ex. (17), haṃsu is nominative because the main verb cāliu is intransitive, whereas in
ex. (18), we find an oblique (ergative) form for yakṣi, because the verb ghātiu is transitive.
(17) isauṃ bhaṇī-kari haṃsu rājā āghau cāliu.
like talk-CVB Hansu king[M]NOM.SG far go.PST.M.SG
’Having said this king Hansu went further’.14th c. (R.G.)
(18) yakṣ-i arjuna ripu bāṃdhī-karī page
Yaksha[M]-OBL.SG Arjuna enemy[M]NOM.SG bind-CVB foot.M.LOC.PL
āṇi ghātiu
come.CVB throw.PST.M.SG
’Yaksha, having bound the enemy named Arjuna, threw him on his feet’. 14th c. (R.G.)
The controller of the converb, however, does not always need to be the same subject as that of the
main verb. Consider, for instance, the following examples:
(19) ara hemū Pāṇīpaṃtha āi derā pariyā
and hemu.M.NOM.SG Panipat come.CVB camp.M.NOM.PL fall.PST.M.PL
‘ And after that Hemu had come to Panipat, the camps were established.’ 16/17th
c. (R.G.)
(20) ti puruṣa raja=nai vacani karī saṃgha=māhi
these man.M.NOM.PL king=OBL speech.F.NOM.SG do.CVB community=in
gayā
go.PST.M.PL
‘These men on hearing the king’s speech (lit. of the king having spoken) went happy to
their community.’16/17th
(R.G.)
(21) tina sahanāna-nūṃ dekha mo=nūṃ khabara paṛasai
that sign=OBL see.CVB I=OBL information be found.PRS.SBJV.3SG
‘Having seen the sign, I would get the information.’ 18th
(R.G.)
Examples (19) and (20) are actually a kind of an absolute construction (i.e. construction not
sharing a subject argument with the main clause) which had been attested in IA throughout its
history but with an inflected participle as the verbal form being in concord with a head noun.
Comparative evidence from other early NIA tongues, e.g. Awadhi (22), (24) or Pahari (Old
Nepali) (23) shows that same subject constraint was not always observed in converbal chain
constructions. Examples (21), (23) and (24) in particular are interesting, since they show that the
subject of the converb is coreferential with the experiencer subject in a non-nominative case of
the main verb (even if it is not overt as in (24)).
Experiencer subjects occupy a special place in the discussion on subjecthood in South Asian
languages. In contemporary NIA they are almost fully endowed with subject behavioural
properties – e.g. they control (but do not undergo) coreferential deletion in conjunction reduction
and converbal chain constructions, control reflexives etc. (for an extensive discussion see Verma
and Mohanan 1991). Our data only confirms that in early NIA dative experiencers had exactly
the same subject properties as in contemporary NIA.
(22) Old Awadhi (J.52.5)
sūra parasa=soṃ bhaeu kirīrā
sun.M.NOM.SG touchstone=ABL become.PST.M.SG play.M.NOM.SG
kirina jāmi upanā naga
ray.F.NOM.SG be born.CVB origin.PST.M.SG precious stone.M.NOM.SG
hīrā
diamond.M.NOM.SG
‘When the play has started from the sun’s touchstone, sunrays having been born, precious
stones diamonds were formed.’
(23) Old Nepali 1398 A.D. (Chalise 2007: 267)
4 kilā bhītra-ko jagā jag-era apnaṃ gar-i
4 boundary inside-GEN cultivated land save-CVB own make.CVB
rāmadāsa pādhyā-lāhi bramavītrā mayā bhai-cha
Ramdas Padhya-OBL grand of land to a Brahmin gift be.PST-3SG.PRS
‘Having saved and made as its own cultivated land within four boundaries the brahmin’s
gift was presented to Ramdas Padhya.’
(24) Old Awadhi (J.31.8)
dekhi rūpa saravara=kara, gai piāsa
see.CVB beauty lake=GEN go.PST.F.SG thirst.F.NOM.SG
au bhūkha
and hunger.F.NOM.SG
‘(Somebody)having seen the beauty of the lake, thirst and hunger are gone.’
Note, however, that the subject marking is sometimes obscured as in ex. (25), because the
ergative case marking gradually disappears in Old Rajasthani, in favour of a nominative-
accusative argument pattern.
(25) sūrajamala doḍa-nai pūraṇamala=nūṃ pāḍiyo
Surajamal.NOM.SG run-CVB Puranmal=ACC knock down.PST.M.SG
‘Surajmal having ran knocked Puranmal down.’ 17th
c. (R.G.)
With finite constructions, we notice that at least up to the 16th
century, there is seemingly
unmotivated variation between marked and unmarked forms of A. Only with the pronominal
forms, A and S6 receive a different, specific marking up till the 18
th century (instrumental/oblique
for A, nominative for S). In Modern Rajasthani, only in the pronominal paradigm one finds
6 S stands for an intransitive subject, A for a transitive subject and O for transitive object, following Dixon 1994: 6.
remnants of a different marking for A and S, nouns are in general in an identical case for A and
S.
Whereas subjects/agents of converbs are restricted to arguments of the main sentence, the
converb can take any direct object/patient. In contrast to the subject, the object of a converb does
not need to be shared with the main verb. This is one of the features which groups converbs with
other non-finite verb forms such as participles and infinitives: they are all more patient- than
agent-oriented. Consider the following examples:
(26) muṃḍa pākhaṃḍika eka=rahaiṃ dekhī+karī 'a-śakunu eu'
shaven.NOM.M.SG ascetic.M.NOM.SG one=ACC see+CVB bad omen
iṇi kāraṇi teha mārāwiwā kāraṇi amhe
this.OBL reason.OBL he.OBL.SG kill.CAUS.INF.OBL reason.OBL we.NOM
mokaḷiyā
send.PST.M.PL
‘Having seen one, a shaven ascetic, ‘a bad omen’ because of that (they) send us to kill
him.’ 14th
c. (R.G.)
(27) mṛga=rahaiṃ mati prayogi chodawī-karī āpaṇau
deer=ACC intelligence.F.NOM.SG use.M.OBL.SG release-CVB own.M.SG
vratu alīka-vacana-parihāra-lakṣaṇu
oath.M.NOM.SG indication of the abandonment of the false speech.M.NOM.SG
akhaṃḍu pratipaḷāi
unbroken save.PRS.3SG
‘Having released the deer by the use of intelligence (the king) is saving his own vow
unbroken by the indication of the abandonment of the false speech.’ 14th
c. (R.G.)
Most Indo-Aryan languages show a particular type of differential object marking, determined by
the factors of animacy and definiteness: animate arguments are marked with an object case
marker (either an oblique case ending or a postpositional marker), whereas inanimate arguments
are only marked when they are definite. Insofar this has been investigated, the pattern of object
marking in New Indo-Aryan is identical for the main verb as well as for the converbal
constructions. The following examples from Hindi illustrate the differential object marking with
converbs:
(28) rām=ko dekh-kar sītā bhag gayī
Rām=ACC see-CVB Sītā.F.NOM.SG ran away go.AUX.PST.F.SG
`Having seen Rām Sītā ran away`
(29) cāy pī-kar madhurῑ akhbār paṛhne
tea.F.NOM.SG drink-CVB Madhuri.F.NOM.SG newspaper reading.INF.OBL
lagī
start.AUX.PST.F.SG
‘Having drunk tea, Madhuri started reading a newspaper.’
In Old Rajasthani, the pattern of the object marking follows the well-known rules of differential
object marking in Indo-Aryan. In our data overtly marked O’s of converbs appear in the text from
the end of the 16th
century (30) and it seems that from this time onwards there is a substantial
increase in the marking of O-arguments, which we can first observe with animate and then within
animate definite arguments. From the 18th
century the marking of the latter group of arguments
seems to become consistent (compare (30) with (21)).
(30) tiṇi edala=nūṃ māri-ara ṭīko liyo
he.OBL.SG Adel=ACC kill-CVB throne.M.NOM.SG take.PST.M.SG
‘He, having killed Adel, took the throne.’ 16/17th
c. (R.G.)
In Early Western Hindi in general, the marking of O gradually grows in consistency. The
following example illustrates that in the earliest stages, even pronominal O’s are not necessarily
marked (31). Even in 19th
century texts of Pahari dialects such as Kumaoni, we find unmarked
pronominal O’s of the converb (32). Nowadays, Pahari dialects often do not mark pronominal
O’s. In these dialects, finite verb constructions do not necessarily show a differential object
marking pattern. This leads to the observation that unmarked O’s are widely attested in
contemporary Pahari dialects belonging to various branches (cf. Stroński 2011). However, more
standardized languages like Kumaoni or Garhwali tend to follow the general IA differential
object marking pattern, and converbs consistently take marked pronominal O’s.
(31) jihi haüṁ gahi chaṇḍiyau
who.OBL.SG I.NOM catch.CVB release.PST.M.SG
‘Who, having caught me, released me.’ (Ch. from Miltner 1995: 50)
(32) unan maiṃ dekhi baṛi rīs ai aur
he.OBL.PL I.NOM.SG see.CVB big anger.F. come.PST.F.SG and
maiṃ=kaṇi mari diyo
I=ACC kill give.AUX.PST.M.SG
‘Having looked at me they became very angry and they killed me.’ (Standard Kumauni)
(Grierson 1916: 172)
O marking in converbal constructions seems to be quite stable in Early Awadhi, already by the
middle of the 16th
century (cf. both nominal and pronominal O in (33) and (34)).
(33) vajra-hiṁ tinakai māri uṛāī
thunderbolt=ACC straw.OBL.M.SG strike.CVB make fly.PRS.3SG
‘Having striken the thunderbolt with a straw he makes it fly.’ (J.6.5)
(34) mohi taji saṁvari jo ohi marasi, kauna
I.OBL abandon.CVB remember.CVB who this.OBL die.PRS.2SG what
lābha tehi hoi
profit you.OBL be.PRS.3SG
‘Having abandon me, you, who remember her, die, what is then the profit for you?’
(J. 209, 10)
If we make a brief comparison with the marking of the arguments in finite constructions and its
diachronic evolution, we notice that the O marking first occurs in the imperfective constructions.
According to Khokhlova (1992: 79), O marking in perfective tenses has only been established at
the turn of the 17th
/18th
century, and it seems likely that this development is an analogical
extension of the O marking in the imperfective. In our corpus, we find more than double the
amount of marked O’s in imperfective constructions than in perfectives. Research has shown that
there is no noticeable difference between the introduction of O marking in the pronominal and
nominal system, the first attestations of the pronominal O marking in the perfective domain are
again around the 18th century (Khokhlova 1992: 79; 1995: 19–20; 2006: 167–168). If we
compare this development with the O-marking in the non-finite constructions, then it seems that
the O-marking with non-finites precedes the O-marking in finite constructions, and certainly the
O-marking in finite perfective constructions. We might conclude that this is caused by the
indifference of converbs to tense/aspect.
4. Semantics of converbs in Old Rajasthani
The main function of converbs in NIA seems to be clause chaining, or, according to some
accounts, subordination. In NIA pure adverbial modification also occurs in converbal form,.
Consider for instance the following example from Hindi, where sitting modifies the action of
talking.
(35) vah baith-kar bolne lagī
she sit-CVB talk.INF.OBL start.PST.F.SG
‘Sitting, she started talking.’
However, often this form is replaced by an adjectival participial, which refers to the subject:
(36) vah baithe hue bolne lagī
she sit.PTCP.PST be.PTCP.PST talk.INF.OBL start.PST.F.SG
‘She, sitting, started talking.’7
Although there are a number of different possible formations of converbs in Old Rajasthani, an
accompanying semantic differentiation is far from being systematic. Nonetheless, some
preliminary observations can be formulated.
4.1. Converbs in light verb constructions
Firstly, converbs are associated with the so-called light verb construction. There has been a vast
literature on the category of ‘light verb’ in IA; however, only a few studies display a diachronic
typological bias. They usually explore possible scenarios of the transition from a converb + V
complex to a light verb, accompanied by various implications pertaining to morphosyntactic and
semantic changes (cf. Hook 1991, 1993; Butt and Lahiri 2013; Slade 2013).
From the earliest Rajasthani sources a light verb construction consists of the main verb (the form
in -ī (-i)), combined with an auxiliary. The auxiliary has lost a part of its lexical meaning and
instead attributes meaning, such as reinforcement, to the meaning of the main verb. One possible
account of the diachronic evolution of such constructions, which are fairly common in Indo-
Aryan, is that the main verb was originally a converb that expressed a modification of the event.
7 In recent typological literature on IA there has been attempt to explain the basic difference of the two non-finite
devices in functional terms. According to Subbarao (2012: 264-272) the converb (on -kar) is subject-oriented
whereas the adverbial participle does not have to be. Our data on Old Rajasthani clearly shows that adverbial
participles are often part of absolute constructions which per se have different subjects, but it is a topic which
requires further study.
However, gradually, by frequent combining of the same lexical pairs, the conjugated verb started
modifying the converb instead. At a certain moment there must have been a functional imbalance
since forms in -ī (-i) were used as both independent converbs and as part of light verbs. This was
perhaps resolved by the introduction of more complex converbal forms. In the example (37) we
can observe that there is a light verb vaḷī gayā ‘they returned’ (which has as its main verb form in
-ī) and it is preceded by the compound form praṇamī-karī ‘having bowed’ (the form is actually a
juxtaposition of two converbs) and this in turn is preceded by a form in -ī (here an independent
converb). It is quite possible that what we observe in early Rajasthani is an attempt to employ
new compound forms as linking devices and the beginning of the grammaticalization of -ī (-i)
forms as part of light verbs. It was however a long process and the forms in -ī (-i) as independent
converbs still occur in the 18th
century. The form in -ī (-i) as a part of a light verb started being
replaced by the root form around the 18th
century (38). At the same time, the root forms
continued to serve as independent converbs (39).
(37) isauṃ bhaṇī praṇamī-karī vaḷī gayā
such say.CVB bow.CVB-do.CVB return.CVB go.PST.M.PL
‘Having said this, they bowed and returned’. 14 c. (R.G.)
(38) so bhāga gayo
s/he.NOM.SG flee.CVB go.PST.M.SG
‘S/He ran.’ 18 c. (R.G.)
(39) so koṭa pāṛa mahala cuṇai chai
he.NOM wall of fort cause to fall.CVB palace build.PRS.3SG be.PRS.3SG
‘After having destroyed the walls of the fort he builds a palace’. 18 c. (R.G.)
4.2. Converbs in clause chaining
The most common function of converbs in Indo-Aryan is quite typically, to modify in an
adverbial way, the event expressed by a main verb. However, the frequent combining of converbs
in one sentence which seem to indicate events which follow-up to each other, is very similar to
the so-called “clause-chaining” function of converbs (Bickel 1998, Haspelmath 1995: 21).8
In our corpus, we find some examples of intricate clause-chaining, in which more than two
converbs follow each other while expressing several events.
(40) sāmuhā āwatā muni=rahaiṃ vāṃdī karī
front come.PTCP.PRS.M.PL monk=to homage.F.NOM.SG do.CVB
mārgu melhī+karī āghau cāliu
way.M.NOM.SG leave+CVB far go.PST.M.SG
‘He paid homage to the monks coming from the front, he left the road and he went further
on.’14th
c. (R.C.)
8We do not enter into the discussion whether converbs are forms of subordination or so-called “cosubordination”.
We follow Bickel (1998: 389) and Johanson (1992) that “”converbs” are simply “subordinate” in the sense of being
“dependent”” on a syntactic level. Semantically though, one can make a difference between converbs that modify the
event expressed by the main verb, and converbs that are non-modifying, but pure clause-chaining. The latter type of
converbs show similarities to Bisang’s (1995) “narrative converbs”, generating purely linear sequences.
(41) namaskāra kari, page lāgi, mātha-i
greeting.M.NOM.SG do.CVB foot.M.OBL.PL touch.CVB head.M-OBL.SG
hātha dirāya-nai, āya-nai śrī rāmacaṃdrajī
hand.M.NOM.SG give-CVB come-CVB honourable Ramacandra
ḍāwā hātha=sūṃ dhanuṣa uṭhāya+liyo
left hand=with bow.M.NOM.SG lift.PST.M.SG+AUX.PST.M.SG
‘The honourable Ramacandra greeted, touched the feet, let the hand on his head, came,
and lifted the bow with his left hand.’ 18th
c. (R.G.)
The last construction shows a combinations of forms on –i and of the forms on –nai. Based
purely on the formation of the converbs in -nai, the assumption could arise that these converbs
indicate a much more adverbial use, since the addition of a postposition seems to emphasize the
non-verbality of the form. The following examples illustrate this, they denote a manner that
specifies the action expressed by the main verb, rather than simply indicating a chain of events.
(42) tārai donūṃ asawāra ghoṛā dauṛarāya-nai gayā .
then both rider.M.NOM.PL horse.M.NOM.PL drive-CVB go.PST.M.SG
‘Then both riders went away, driving the horses.’18th
c. (R.G.)
However, such relatively clear occurrences of adverbial modification are few in our corpus. The
dominant meaning of converbs is as clause chaining mechanisms. For instance, the following
examples are almost impossible to interpret in another way than as clause chaining:
(43) āya-nai dilī lī
come-CVB Delhi.F seize.PST.F.SG
‘He came and seized Delhi.’ 16/17th
c. (R.G.)
(44) vīkānera vaḷe rāwa kalyāṇamala āi rāja virājaṇa
Bikaner then king.M.NOM.SG Kalyanmala come.CVB throne sit.INF
lāgā
begin.PST.M.SG
‘Then, king Kalyanmal came to Bikaner and began sitting on the throne.’16/17th
c. (R.G.)
Converbs ending on the suffix –ara, again, are often not used as simple adverbial modification,
but rather as a subordinated clause that expresses an action going on before the main verb.
(45) akabara=rai sāthi phoja de-ara kalanora=nūṃ melhi-ara
Akbar=with army.F.NOM.SG give-CVB Kalanur=ACC send-CVB
pātisāha hamāūṃ dilī āyo.
king.M.NOM.SG Humayun.NOM.SG Delhi come.PST.M.SG.
‘Humayun gave the army with Akbar and sent him to Kalanur, and came to Delhi.’16/17th
c. (R.G.)
Interpreting the data, we are unable to observe discrete semantic differences between the different
forms of converbs. A multivariate statistical analysis performed on our data should be able to
trace tendencies, which is a future undertaking.
5. Conclusion
In this article, we have focused on the morphology, syntax and semantics of converbs in Old
Rajasthani. We have placed them in a wider framework of Indo-Aryan, also selecting some
features which are cross-linguistically attributed to the category of converbs.
There are a number of possible converb formations in Old Rajasthani; however, they are related
to particular periods. The oldest forms are the most resilient, i.e. ī and i, which seem to be
nominal oblique endings transferred to the verbal root. The morphology is illustrative of the dual
nature of converbs, hinging between verbality and nominality. The forms which are constructed
with the multifunctional case suffix nai (identical with the dative/accusative postposition) make
this even more clear. Further, we found traces of the predecessor of the Hindi converbal suffix
kar, apparently deriving from the converb kari that has grammaticalized from a form on its own
to a marker of other converbs, in a sort of unconventional light verb construction. The suffix –ara
was also used, showing a relation between Old Rajasthani and the Pahari languages.
On the syntactic level, we focused on the same-subject constraint, and showed that this was
already in place in Old Rajasthani. We found evidence of differential object marking in converbal
constructions from the turn of the 16th
and 17th
centuries onwards, demonstrating that the first
occurrences of a marked object of a converb must have appeared together with their occurrences
in imperfective constructions. Differential object marking only appeared later with finite
perfective constructions. This different pattern of object marking of perfective finite verbs and
converbs might lead to questioning whether converbs are indeed perfective in nature, as Davison
(1981) assumes. However, it is also possible that their overall stricter nominality motivates the
marking of their animate object arguments.
On a semantic level, we tried to connect converbal forms with particular uses. Apart from
observing that light verb constructions are limited to converbal forms on ī /i and the root forms,
on the basis of our qualitative interpretation of the data, we could not find any convincing
evidence of typical uses of the different forms.
In the future, we plan to extend our corpus to Awadhi and Braj, and extract statistical analyses of
it. The analysis of this Old Rajasthani data is however a promising start in our endeavour to
describe and understand converbs on all levels in Early NIA.
References
Baruah, P.N. Dutta. 1980. An intensive course in Assamese. Mysore: CIIL.
Bickel, Balthasar & Yogendra P. Yadav. 2000. A fresh look at grammatical relations in Indo-
Aryan. Lingua 110: 343-373.
Bickel, Balthasar. 1998. Review article of ‘Converbs in cross-linguistic perspective’. Linguistic
typology 2: 381-397.
Bisang, Walter. 1995. Verb serialization and converbs – differences and similarities. In Martin
Haspelmath & Ekkehard König (eds.). Converbs in Cross-linguistic perspective. Berlin: Mouton
de Gruyter. 137-188.
Bloch, Jules. 1965. Indo-Aryan. From the Vedas to modern times. Paris: Librairie D'amerique Et
D'orient Adrien – Maisonneuve.
Chālise, Bidurkumar. 2006. Aitihāsika abhilekhakā ādhāramā nepālī bhāṣākā vyākaraṇa tatvamā
kramika vikāsako adhyayana. [A study on the stadial development in the grammar of the Nepali
language based on the historical inscriptions]. Unpublished PhD thesis. Kathmandu: Tribhuvan
University.
Davison, Alice. 1981. Syntactic and semantic indeterminacy resolved: A mostly pragmatic
analysis for the Hindi conjunctive participle. In Peter Cole (ed.), Radical pragmatics. New York:
Academic Press, 101-128.
Haspelmath, Martin. 1995. The converb as a cross-linguistically valid category. In Martin
Haspelmath & Ekkehard König (eds.). Converbs in cross-linguistic perspective. Berlin: Mouton
de Gruyter. 1-56.
Hutt, Michael. 1997. Modern literary Nepali. An introductory reader. New Delhi: Oxford
University Press.
Jaworski, Rafał. 2015. IATagger. http://rjawor.vm.wmi.amu.edu.pl/tagging/
Johanson, Lars. 1992. Periodische Kettensätze im Türkischen. Wiener Zeitschrift für die Kunde
des Morgenlandes. 82: 201-211.
Kachru, Yamuna. 1981. On the syntax, semantics and pragmatics of the conjunctive participle in
Hindi-Urdu. Studies in the linguistic sciences 11(2): 35-49.
Khokhlova, Ludmila V. 1992. Trends in the development of ergativity in New Indo-Aryan.
Osmania Papers in Linguistics 18, 71–89.
Khokhlova, Ludmila V. 1995. The Development of patient-oriented constructions in late Western
NIA Languages. Osmania Papers in Linguistics 21, 15–51.
Khokhlova, Ludmila V., 2006. Sintaktičeskaja evolucija zapadnych novoindijskich jazykov v 15–
20 vv. [Syntactic evolution of Western New Indo-Aryan languages in 15-20 c.]. In: Dybo (et al.
eds.) Aspekty komparativistiki [Aspects of comparative studies]. Moskva: Rosijskij
Gosudarstvennyj Gumanitarnyj Universitet (Orientalia et Classica: Trudy Instituta Vostočnych
Kultur i Antičnosti: Vypusk VIII). 151–186.
Masica, Colin. 1976. Defining a linguistic area. South Asia. Chicago: University of Chicago
Press.
Miltner, Vladimir. 1995. Old Hindi Reader. Prague: Karolinum – Charles University Press.
Pokharel, Bālkr̥ṣṇa. 1964. Pānca śaya varṣa [Five hundred years]. Lalitpur: Jagadambāprakāśan.
Reinöhl, Uta. 2016. A single origin of Indo-European primary adpositions? Unveiling the Indo-
Aryan branch-off. Diachronica 33(1): 95-130.
Saksena, Baburam.1971 [1937]. Evolution of Awadhi. Delhi: Motilal Banarsidass.
Stroński, Krzysztof. 2009. On the origin of the ergative postposition in Hindi. Rocznik
Orientalistyczny LXII/1: 175-186.
Stroński, Krzysztof. 2011. Synchronic and diachronic aspect of erativity in Indo-Aryan.Poznań:
Wydawnictwo Naukowe UAM.
Stroński, Krzysztof & Tokaj, Joanna. 2015. The diachrony of cosubordination – lessons from
Indo-Aryan. Paper presented at the 31st South Asian Languages Analysis Roundtable (SALA-
31), Lancaster 14-16 May 2015. Extended abstract available at: http://ucrel.lancs.ac.uk/sala-
31/doc/ABSTRACTBOOK maincontent.pdf.
Subbarao, Kurumuri V. 2012. South Asian languages: a syntactic typology. Cambridge:
Cambridge University Press.
Śukla, Prabhākar. 1965. Jāyasī kī bhāṣā [The language of Jayasi]. Lakhnaū: Lakhnaū
Viśvavidyālay.
Tessitori, Luigi P. 1914–16. Notes on the Grammar of the Old Western Rajasthani with Special
Reference to Apabhramsa and to Gujarati and Marwari. Indian Antiquary 43:5.
Tikkanen, Bertil. 1995. Burushaski converbs in their South and Central Asian areal context. In
Martin Haspelmath & Ekkehard König (eds.). Converrbs in cross-linguistic perspective. Berlin:
Mouton de Gruyter. 487-528.
Verbeke, Saartje. 2013. Alignment and ergativity in the New Indo-Aryan languages. Berlin:
Mouton de Gruyter.
Wallace, William. 1982. The evolution of ergative syntax in Nepali. Studies in the linguistic
sciences 12(2): 147-211.